No matter which part of the NHS you are in, you will face the same fundamental challenge: to deliver outstanding healthcare while staying within financial boundaries which are finite and shrinking. Tough decisions need to be made.

There is increasing pressure to make these decisions immediately and this perceived urgency can easily tempt us to skip the evidence and evaluation stage of the decision-making process. However, planning is fundamental to a successful outcome.

This article explains when and how predictive modelling can be used to ease the burden of this planning stage, giving you confidence that your quick decisions will be the right decisions.

When should predictive modelling be used?

When the decision is open to influence

Predictive modelling adds most value when it is used to inform the decision making process, either making the initial decision or for a re-evaluation, but not simply to justify a declared position.

When simple logic will not suffice

Some decisions can be governed by simple logic e.g. balancing the cost of an energy-saving investment against the expected annual cash saving. However, the complexities and interactions of healthcare services mean that the overall impact of local changes can often be counter-intuitive. A properly built predictive model can address this complexity and highlight the full impact of decisions.

When, you need to deliver against a plan

Predictive modelling will quantify the link between actions (or investment) and outcomes so you can be confident that your plans will enable you to reach your targets. Or, it will enable you to change your plans accordingly, before the money is spent.

When modelling is more timely or cost-effective than trialling

The alternatives to modelling should always be considered.  Sometimes it is more effective to trial a change and monitor the benefits than to invest in some modelling.  However, even when trialling is required, models can usefully rule out options that would either fail to work or would not bring sufficient benefits, thereby saving both time and money.

Why is it worth it?

If you decide to develop a predictive model, it has the potential to transform your decision making processes. 

It will give you a clear view of the big picture

Predictive models offer a balance between simplicity and reality, allowing you to better understand the underlying features of a complex, integrated system of care. This gives you the ability to assess how changes to one part affect the whole.

It will enhance your ability to quickly make the right decisions

Depending on how the model has been designed, it can be used in any or all of the following situations

  1. To demonstrate the link between a proposed cut and the deterioration in service levels that will ensue; then to look for the changes that can deliver cost savings as well as improve levels of care.
  2. To plan for the future by predicting the pressure that will be exerted on existing resources and infrastructure by a change in demographics such as an ageing population, or by an increasing prevalence of a particular condition.  Options can then be identified to alleviate that pressure.
  3. To choose between a range of possible options by understanding the impact they will each have on your key target measures whether financial, patient care, corporate responsibility or other measures.

What to avoid

Locking your modellers away in a darkened room

It is important to incorporate local expertise into all aspects of the model-building process. You know best how your care system operates currently and what your plans are to change it, so make sure your modellers are engaged with you or your team on a day-to-day basis. This will ensure that the model is based on useful and appropriate data, and is a fair reflection of the real world.

Do-it-yourself modelling

Effective predictive modelling is a specialist activity and poorly conducted analysis can be worse than no analysis at all. Unless you have qualified individuals within your team you should seek suitable experts to guide the process. Skills transfer should be an important part of any external modelling contract, with a view to using the model internally and potentially building future models in-house.

Failure to properly test your model

Every model has its limits and it is vital to understand the importance of incorporating the best data and the most valued expert opinions.  The models should be thoroughly tested, not only for functionality but also for their effectiveness at representing reality – your local knowledge is central to this testing process.

Applying the results blindly

A predictive model is an aid to decision making, but it shouldn’t make the decisions for you.  Use your model wisely and it will be a great asset, but do not ignore your own knowledge and experience.

Top Modelling Tips

And finally, some general advice that will stand you in good stead for whatever type of modelling you need

  1. Be clear about the decisions you want the model to support
  2. Put the best available data into the model
  3. Involve local experts in all aspects of the model development (Design, Build, Test, Use )
  4. Let your model support decisions, not make them
  5. Build flexibility into your model so it can be re-used

Predictive modelling has the potential to underpin your thinking in a constantly changing and challenging environment, enabling you to make the right decisions with confidence.